Design an agent to fly a quadcopter, and then train it using a reinforcement learning algorithm of your choice!
Try to apply the techniques you have learnt, but also feel free to come up with innovative ideas and test them.
Take a look at the files in the directory to better understand the structure of the project.
task.py: Define your task (environment) in this file.agents/: Folder containing reinforcement learning agents.policy_search.py: A sample agent has been provided here.agent.py: Develop your agent here.physics_sim.py: This file contains the simulator for the quadcopter. DO NOT MODIFY THIS FILE.For this project, you will define your own task in task.py. Although we have provided a example task to get you started, you are encouraged to change it. Later in this notebook, you will learn more about how to amend this file.
You will also design a reinforcement learning agent in agent.py to complete your chosen task.
You are welcome to create any additional files to help you to organize your code. For instance, you may find it useful to define a model.py file defining any needed neural network architectures.
We provide a sample agent in the code cell below to show you how to use the sim to control the quadcopter. This agent is even simpler than the sample agent that you'll examine (in agents/policy_search.py) later in this notebook!
The agent controls the quadcopter by setting the revolutions per second on each of its four rotors. The provided agent in the Basic_Agent class below always selects a random action for each of the four rotors. These four speeds are returned by the act method as a list of four floating-point numbers.
For this project, the agent that you will implement in agents/agent.py will have a far more intelligent method for selecting actions!
%%javascript
IPython.OutputArea.prototype._should_scroll = function(lines) {
return false;
}
// To disable scroll output in the notebook.
import random
class Basic_Agent():
def __init__(self, task):
self.task = task
def act(self):
new_thrust = random.gauss(450., 25.)
return [new_thrust + random.gauss(0., 1.) for x in range(4)]
Run the code cell below to have the agent select actions to control the quadcopter.
Feel free to change the provided values of runtime, init_pose, init_velocities, and init_angle_velocities below to change the starting conditions of the quadcopter.
The labels list below annotates statistics that are saved while running the simulation. All of this information is saved in a text file data.txt and stored in the dictionary results.
%load_ext autoreload
%autoreload 2
import csv
import numpy as np
from task import Task
# Modify the values below to give the quadcopter a different starting position.
runtime = 100. # time limit of the episode
init_pose = np.array([0., 0., 10., 0., 0., 0.]) # initial pose
init_velocities = np.array([0., 0., 0.]) # initial velocities
init_angle_velocities = np.array([0., 0., 0.]) # initial angle velocities
file_output = 'data.txt' # file name for saved results
# Setup
task = Task(init_pose, init_velocities, init_angle_velocities, runtime)
agent = Basic_Agent(task)
done = False
labels = ['time', 'x', 'y', 'z', 'phi', 'theta', 'psi', 'x_velocity',
'y_velocity', 'z_velocity', 'phi_velocity', 'theta_velocity',
'psi_velocity', 'rotor_speed1', 'rotor_speed2', 'rotor_speed3', 'rotor_speed4']
results = {x : [] for x in labels}
# Run the simulation, and save the results.
with open(file_output, 'w') as csvfile:
writer = csv.writer(csvfile)
writer.writerow(labels)
while True:
rotor_speeds = agent.act()
_, _, done = task.step(rotor_speeds)
to_write = [task.sim.time] + list(task.sim.pose) + list(task.sim.v) + list(task.sim.angular_v) + list(rotor_speeds)
for ii in range(len(labels)):
results[labels[ii]].append(to_write[ii])
writer.writerow(to_write)
if done:
break
Run the code cell below to visualize how the position of the quadcopter evolved during the simulation.
import matplotlib.pyplot as plt
%matplotlib inline
plt.plot(results['time'], results['x'], label='x')
plt.plot(results['time'], results['y'], label='y')
plt.plot(results['time'], results['z'], label='z')
plt.legend()
_ = plt.ylim()
The next code cell visualizes the velocity of the quadcopter.
plt.plot(results['time'], results['x_velocity'], label='x_hat')
plt.plot(results['time'], results['y_velocity'], label='y_hat')
plt.plot(results['time'], results['z_velocity'], label='z_hat')
plt.legend()
_ = plt.ylim()
Next, you can plot the Euler angles (the rotation of the quadcopter over the $x$-, $y$-, and $z$-axes),
plt.plot(results['time'], results['phi'], label='phi')
plt.plot(results['time'], results['theta'], label='theta')
plt.plot(results['time'], results['psi'], label='psi')
plt.legend()
_ = plt.ylim()
before plotting the velocities (in radians per second) corresponding to each of the Euler angles.
plt.plot(results['time'], results['phi_velocity'], label='phi_velocity')
plt.plot(results['time'], results['theta_velocity'], label='theta_velocity')
plt.plot(results['time'], results['psi_velocity'], label='psi_velocity')
plt.legend()
_ = plt.ylim()
Finally, you can use the code cell below to print the agent's choice of actions.
plt.plot(results['time'], results['rotor_speed1'], label='Rotor 1 revolutions / second')
plt.plot(results['time'], results['rotor_speed2'], label='Rotor 2 revolutions / second')
plt.plot(results['time'], results['rotor_speed3'], label='Rotor 3 revolutions / second')
plt.plot(results['time'], results['rotor_speed4'], label='Rotor 4 revolutions / second')
plt.legend()
_ = plt.ylim()
When specifying a task, you will derive the environment state from the simulator. Run the code cell below to print the values of the following variables at the end of the simulation:
task.sim.pose (the position of the quadcopter in ($x,y,z$) dimensions and the Euler angles),task.sim.v (the velocity of the quadcopter in ($x,y,z$) dimensions), andtask.sim.angular_v (radians/second for each of the three Euler angles).# the pose, velocity, and angular velocity of the quadcopter at the end of the episode
print(task.sim.pose)
print(task.sim.v)
print(task.sim.angular_v)
In the sample task in task.py, we use the 6-dimensional pose of the quadcopter to construct the state of the environment at each timestep. However, when amending the task for your purposes, you are welcome to expand the size of the state vector by including the velocity information. You can use any combination of the pose, velocity, and angular velocity - feel free to tinker here, and construct the state to suit your task.
A sample task has been provided for you in task.py. Open this file in a new window now.
The __init__() method is used to initialize several variables that are needed to specify the task.
PhysicsSim class (from physics_sim.py). action_repeats timesteps. If you are not familiar with action repeats, please read the Results section in the DDPG paper.state_size), we must take action repeats into account. action_size=4). You can set the minimum (action_low) and maximum (action_high) values of each entry here.The reset() method resets the simulator. The agent should call this method every time the episode ends. You can see an example of this in the code cell below.
The step() method is perhaps the most important. It accepts the agent's choice of action rotor_speeds, which is used to prepare the next state to pass on to the agent. Then, the reward is computed from get_reward(). The episode is considered done if the time limit has been exceeded, or the quadcopter has travelled outside of the bounds of the simulation.
In the next section, you will learn how to test the performance of an agent on this task.
The sample agent given in agents/policy_search.py uses a very simplistic linear policy to directly compute the action vector as a dot product of the state vector and a matrix of weights. Then, it randomly perturbs the parameters by adding some Gaussian noise, to produce a different policy. Based on the average reward obtained in each episode (score), it keeps track of the best set of parameters found so far, how the score is changing, and accordingly tweaks a scaling factor to widen or tighten the noise.
Run the code cell below to see how the agent performs on the sample task.
import sys
import pandas as pd
from agents.policy_search import PolicySearch_Agent
from task import Task
num_episodes = 1000
target_pos = np.array([0., 0., 10.])
task = Task(target_pos=target_pos)
agent = PolicySearch_Agent(task)
for i_episode in range(1, num_episodes+1):
state = agent.reset_episode() # start a new episode
while True:
action = agent.act(state)
next_state, reward, done = task.step(action)
agent.step(reward, done)
state = next_state
if done:
print("\rEpisode = {:4d}, score = {:7.3f} (best = {:7.3f}), noise_scale = {}".format(
i_episode, agent.score, agent.best_score, agent.noise_scale), end="") # [debug]
break
sys.stdout.flush()
This agent should perform very poorly on this task. And that's where you come in!
Amend task.py to specify a task of your choosing. If you're unsure what kind of task to specify, you may like to teach your quadcopter to takeoff, hover in place, land softly, or reach a target pose.
After specifying your task, use the sample agent in agents/policy_search.py as a template to define your own agent in agents/agent.py. You can borrow whatever you need from the sample agent, including ideas on how you might modularize your code (using helper methods like act(), learn(), reset_episode(), etc.).
Note that it is highly unlikely that the first agent and task that you specify will learn well. You will likely have to tweak various hyperparameters and the reward function for your task until you arrive at reasonably good behavior.
As you develop your agent, it's important to keep an eye on how it's performing. Use the code above as inspiration to build in a mechanism to log/save the total rewards obtained in each episode to file. If the episode rewards are gradually increasing, this is an indication that your agent is learning.
# Imports
import sys
import numpy as np
from agents.agent import DDPG
from new_task import Task
import matplotlib as mpl
import matplotlib.pyplot as plt
from mpl_toolkits.mplot3d import Axes3D
import pandas as pd
# Modify the values below to give the quadcopter a different starting position.
runtime = 5 # Time limit of the episode.
init_pose = np.array([0., 0., 100., 0., 0., 0.]) # Initial position of the quadcopter.
init_velocities = np.array([0., 0., 1.]) # Initial velocity of the quadcopter.
init_angle_velocities = np.array([0., 0., 0.]) # Initial angular velocities of the quadcopter.
target_pos = np.array([0., 0., 200.]) # Target position to reach.
# Labels for taking logs.
labels = ['episode', 'time', 'x', 'y', 'z', 'phi', 'theta', 'psi',
'x_velocity', 'y_velocity', 'z_velocity', 'phi_velocity',
'theta_velocity', 'psi_velocity', 'rotor_speed1', 'rotor_speed2',
'rotor_speed3', 'rotor_speed4', 'rewards', 'score', 'max_score', 'max_avg_score']
# Setup:
# Creating a Task object with the specified target position.
task = Task(init_pose, init_velocities, init_angle_velocities, runtime, target_pos)
# Creating an instance of the agent.
agent = DDPG(task)
# Specifying the number of episodes
num_episodes = 1000
# List for maintaining the score of each episode.
scores = []
# Max average score (over 100 episodes).
max_avg_score = -np.inf
# Max score achieved during training.
max_score = -np.inf
# Dictionary for taking logs.
logs = {}
for i_episode in range(1, num_episodes+1):
# Initialize episode.
state = agent.reset_episode() # Start a new episode.
total_reward = 0
results = {x : [] for x in labels} # For storing result after each episode.
while True:
results['time'].append(task.sim.time)
# Take action, evaluate reward and next state, update policy function.
action = agent.act(state)
next_state, reward, done = task.step(action)
results['rewards'].append(reward)
agent.step(action, reward, next_state, done)
total_reward += reward
# Storing the value of quadcopter's variables(position, velocity, angular velocity) after each timestep in "results".
quadcopter_state = list(task.sim.pose) + list(task.sim.v) + list(task.sim.angular_v)
for i, s in enumerate(quadcopter_state):
results[labels[i+2]].append(s)
for a in range(len(action)):
results[labels[14+a]].append(action[a])
state = next_state
if done:
max_score = max(total_reward, max_score)
print("\rEpisode = {:4d}, score = {:7.3f} (max_score = {:7.3f}, max_avg_score = {:7.3f}), last_position = ({:5.1f},{:5.1f},{:5.1f}))"
.format(i_episode, total_reward, max_score, max_avg_score, task.sim.pose[0], task.sim.pose[1], task.sim.pose[2]), end="")
# Storing result after each episode.
results['episode'] = i_episode
results['score'] = total_reward
results['max_score'] = max_score
results['max_avg_score'] = max_avg_score
# Maintaining logs of each episode's result.
logs[i_episode] = results
break
sys.stdout.flush()
# Save final score
scores.append(total_reward)
# Update maximum average score attained over 100 consecutive trials.
if len(scores) > 100:
avg_score = np.mean(scores[-100:])
max_avg_score = max(avg_score, max_avg_score)
Once you are satisfied with your performance, plot the episode rewards, either from a single run, or averaged over multiple runs.
# Making a dataframe from the logs and saving it into a .csv file.
logs = pd.DataFrame.from_dict(data=logs, orient='index')
logs.to_csv('logs',index=False)
# Plotting the scores vs episode graph.
plt.figure(figsize=(16,4))
plt.plot(logs['episode'], logs['score'])
plt.title("Scores vs Episodes")
plt.legend()
# Finding the episode in which the agent achieves the maximum score.
max_score_episode = np.argmax(logs["score"])
logs.iloc[max_score_episode-1]
# Visualization
import importlib
import visualize
import imageio
from IPython.display import Image
importlib.reload(visualize)
# Utility function to show drone path in a specified episode.
def plot_drone_path(logs=logs, episode = num_episodes, target_pos = target_pos, x_lim = (-50, 50),
y_lim = (-50, 50), z_lim = (0, 250), figsize=(10, 5), t_file_path=None, r_file_path=None):
qd = visualize.Quadrotor(x = logs.iloc[episode-1]['x'][0], y = logs.iloc[episode-1]['y'][0],
z = logs.iloc[episode-1]['z'][0], roll = logs.iloc[episode-1]['phi'][0],
pitch = logs.iloc[episode-1]['theta'][0], yaw = logs.iloc[episode-1]['psi'][0],
target_pos = target_pos, x_lim=x_lim, y_lim=y_lim, z_lim=z_lim, figsize=figsize, size=10)
t_images = []
r_images = []
for t in range(len(logs.iloc[episode-1]['time'])):
x = logs.iloc[episode-1]['x'][t]
y = logs.iloc[episode-1]['y'][t]
z = logs.iloc[episode-1]['z'][t]
roll = logs.iloc[episode-1]['phi'][t]
pitch = logs.iloc[episode-1]['theta'][t]
yaw = logs.iloc[episode-1]['psi'][t]
reward = logs.iloc[episode-1]['rewards'][t]
time = logs.iloc[episode-1]['time'][t]
t_image, r_image = qd.update_pose(x=x, y=y, z=z, roll=roll, pitch=pitch, yaw=yaw, reward=reward, time=time)
t_images.append(t_image)
r_images.append(r_image)
if t_file_path is None:
t_file_path = './drone_path.gif'
if r_file_path is None:
r_file_path = './RewardVsTime.gif'
# Save animated gif
imageio.mimsave(t_file_path, t_images, fps=5)
imageio.mimsave(r_file_path, r_images, fps=5)
# Show animated gif for live drone trajectory
with open(t_file_path,'rb') as f:
display(Image(data=f.read(), format='png'))
# Show animated gif for reward versus time
with open(r_file_path,'rb') as f:
display(Image(data=f.read(), format='png'))
fig = plt.figure(figsize=figsize)
ax = fig.gca(projection='3d')
x = logs.iloc[episode-1]['x']
y = logs.iloc[episode-1]['y']
z = logs.iloc[episode-1]['z']
ax.plot3D(x, y, z, label='Drone path', alpha = 0.75, color = 'red')
ax.set_xlim(x_lim)
ax.set_ylim(y_lim)
ax.set_zlim(z_lim)
ax.scatter(target_pos[0], target_pos[1], target_pos[2], label='Target position', color = 'green')
ax.set_title("Drone Trajectory")
ax.legend()
# Plotting the drone path and reward for the max score episode.
plot_drone_path(episode = max_score_episode)
Question 1: Describe the task that you specified in task.py. How did you design the reward function?
Answer: I specified the task to be a takeoff mission, that is the task would be to reach some defined height from an initial position (point from where the drone starts its journey). Mission : (x, y, h) -> (x, y, H); h-> Initial Height; H -> Target Height.
The reward function that I defined takes into account many different scores to compute the reward value, these are:
The final reward is calculated by summing up all these scores and then scaling it by using tanh function which gives a very nice non-linear reward range.
Question 2: Discuss your agent briefly, using the following questions as a guide:
Answer:
The final choice was as follows:
Update parameters->
gamma = 0.99
tau = 0.02
Replay buffer parameters->
buffer size = 100000
batch size = 64
Noise parameters->
mu = 0.05
theta = 0.05
sigma = 0.05
I used the suggested neural network architecture, I just changed the size and types of the hidden layers.
For the actor:
Dense(size=256, l2_regularizer(0.001))
Batch normalization layer
Relu activation layer
Dense(size=128, l2_regularizer(0.001))
Batch Normalization layer
Relu activation layer
For the critic:
state pathway->
Dense(size=256, l2_regularizer(0.001))
Batch noarmalization layer
Relu activation layer
Dense(size=256, l2_regularizer(0.001))
action pathway->
Dense(size=256, l2_regularizer(0.001))
Batch noarmalization layer
Relu activation layer
I used batch normalization and L2 regularization to optimize the network.
Question 3: Using the episode rewards plot, discuss how the agent learned over time.
Answer:
Question 4: Briefly summarize your experience working on this project. You can use the following prompts for ideas.
Answer: